HOW CHANGES IN ENTRY

REQUIREMENTS ALTER THE

TEACHER WORKFORCE AND

AFFECT STUDENT ACHIEVEMENT

抽象的
We are in the midst of what amounts to a national experi-
ment in how best to attract, prepare, and retain teachers,
particularly for high-poverty urban schools. 使用数据
on students and teachers in grades 3–8, this study as-
sesses the effects of pathways into teaching in New York
City on the teacher workforce and on student achieve-
蒙特. We ask whether teachers who enter through new
routes, with reduced coursework prior to teaching, 是
more or less effective at improving student achievement.
When compared to teachers who completed a university-
based teacher education program, teachers with reduced
coursework prior to entry often provide smaller initial
gains in both mathematics and English language arts.
Most differences disappear as the cohort matures, 和
many of the differences are not large in magnitude, typi-
卡莉 2 到 5 percent of a standard deviation. The variation
in effectiveness within pathways is far greater than the
average differences between pathways.

Donald Boyd

Center for Policy Research

University at Albany

Albany, 纽约 12222

boydd@rockinst.org

Pamela Grossman

School of Education

斯坦福大学

斯坦福大学, CA 94305

pamg@stanford.edu

Hamilton Lankford

Economics Department

University at Albany

Albany, 纽约 12222

hamp@albany.edu

Susanna Loeb

School of Education

斯坦福大学

斯坦福大学, CA 94305

sloeb@stanford.edu

James Wyckoff

(corresponding author)

Rockefeller College of Public

Affairs and Policy University

at Albany

Albany, 纽约 12222

wyckoff@albany.edu

176

C(西德:1) 2006 American Education Finance Association

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

1. 介绍
Demographic changes and new policies, such as class size reduction, 是
increasing both the need for new teachers and the need for a greater under-
standing of how to prepare effective teachers. In New York City, 例如,
retirements and demand for teachers are straining the ability of existing teacher
preparation programs to produce sufﬁcient numbers of teachers. 此外,
new standards for high achievement by all students require newly entering
teachers to be more skilled than in the past. In low-performing schools with
high proportions of poor and nonwhite students, the qualiﬁcations of teachers
are already substantially worse than in better-performing urban and suburban
学校 (看, 例如, Lankford, Loeb, and Wyckoff 2002). As the demand
for high-quality teachers increases, these disparities are likely to worsen as
schools with better working conditions and higher salaries bid away the better-
qualiﬁed teachers from difﬁcult-to-staff schools. 此外, there is evidence
that a substantial majority of teachers take their ﬁrst job very close to where they
grew up and that urban areas hire more teachers than individuals who choose
education as a profession from those areas, thus having to hire teachers from
other areas, further disadvantaging urban schools (Boyd et al. 2005). 许多
urban districts have begun to rely on new teacher preparation programs that
greatly reduce the requirements for coursework and experiences in schools
prior to becoming a teacher but provide supports and additional coursework
during the ﬁrst years of teaching. Using data on students and teachers in grades
3–8 in New York City, this study assesses the effects of such programs on the
teacher workforce and on student achievement. We ask whether teachers who
enter through these new routes are more or less effective at improving student
achievement than other teachers and whether the presence of these alternative
pathways affects the composition of New York City’s teaching workforce.

Recent research has documented the importance of teachers to student
achievement (看, 例如, Rivkin, Hanushek, and Kain 2005; Rockoff
2004; Sanders and Horn 1994; Sanders and Rivers 1996). Each of these stud-
ies documents the improvement of student achievement with increases in
teacher experience during the ﬁrst three to ﬁve years of experience, with virtu-
ally no additional gains for experience beyond ﬁve years. Many other factors,
including teachers’ verbal ability and subject matter preparation, contribute to
teacher effectiveness (Ehrenberg and Brewer 1995; Monk 1994). Schools may
also affect teacher effectiveness through, 例如, 资源, 管理-
tive leadership, and/or curriculum. There is some evidence that professional
education can improve teachers’ abilities in the classroom (National Research
理事会 2001; Brown and Borko 1992; Garet et al. 2001; Loucks-Horsley and
Matsumoto 1999; Monk 1994). This study focuses speciﬁcally on pathways
into teaching.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

177

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

At the same time that schools face an increased need for new teachers,
researchers and policy makers are engaged in debates over the best way to pre-
pare teachers. While some argue that easing entry into teaching is the best way
to attract strong candidates (我们. 教育部 2002; Finn and
Madigan 2001), others argue that investing in high-quality teacher prepara-
tion will better serve schools and students (National Commission on Teaching
and America’s Future 1996). Most agree, 然而, that we lack a strong re-
search basis for understanding how best to structure preparation requirements
for teaching (看, 例如, Wilson, Floden, and Ferrini-Mundy 2002).
While there are many studies of how aspects of preparation affect teacher
信仰, 知识, and practices,1 much of the research is limited in scope,
focuses on inputs rather than outcomes, or employs case study methodologies
from which it is difﬁcult to determine causal relationships or generalize to
other populations. 因此, discussions about effective teacher preparation
could beneﬁt from more methodologically sophisticated research to inform the
辩论, especially as they pertain to preparing teachers for urban settings
(see Cochran-Smith and Zeichner 2005).

Research on the effects of teacher preparation programs with reduced
requirements prior to teaching is scarce—not surprisingly, given the relatively
recent emergence of these programs. Some evidence, 然而, suggests that
such programs increase the supply of teachers and change the composition
of the teacher workforce (see McKibbon 1998; 沉 1999). 然而, to our
知识, there is no work that assesses the effectiveness of teachers from
large-scale programs at improving students’ academic achievement.

The most relevant studies of student achievement effects come from three
studies of Teach for America (TFA), a national program that recruits college
seniors to teach in low-income communities for two years. 雷蒙德, 弗莱彻,
and Luque (2001) ﬁnd that, in Houston, students of TFA teachers had higher
test score gains in mathematics than did students of other teachers, 许多
of whom were not certiﬁed. Darling-Hammond et al. (2005) use the same
data and ﬁnd mixed effects of TFA teachers on student test score gains, 和
students of TFA teachers performing relatively better in tests of mathematics
than English language arts (ELA). Decker, Mayer, and Glazerman (2004) 是
able to compare teachers within the same grade and school in an experimental
format that randomly assigned teachers to students. Using a sample of seven-
teen schools from Baltimore, 芝加哥, 天使们, Houston, New Orleans,
and the Mississippi Delta, they ﬁnd that TFA teachers were more effective at

For relevant work see Ball and Cohen 1999, Carnegie Forum on Education and the Economy 1986,
Darling-Hammond 2000, Darling-Hammond and McLaughlin 1999, Feiman-Nemser 1983, 1990,
Goodlad 1990, and Holmes Group 1986.

178

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

improving student math scores than were other teachers, though there were
no statistically signiﬁcant differences in reading scores.

The TFA results may or may not mirror the effects of other programs
that have reduced preservice preparation aimed at increasing the supply of
teachers to urban schools. TFA has brought a substantial number of teach-
ers into schools, 大约 12,000 自从 1990, and the applicant pool
has been growing. 之间 2000 和 2003 the applicant pool grew from
4,068 到 15,706, and the number of new TFA teachers increased from 868
到 1,656. 然而, it provides many fewer teachers to urban schools than
do the new programs designed speciﬁcally to increase the supply of teachers
(Decker, Mayer, and Glazerman 2004). 例如, 在 2004, the New York
City Teaching Fellows program, an alternative route into teaching designed
speciﬁcally to staff New York City schools, provided almost 2,500 new teach-
ers to New York City, while TFA provided fewer than 350. Because they are
so large, these new pathways may drastically change the composition of the
teacher workforce in city schools. We are not aware of any studies that assess
the effectiveness of these large-scale programs.

2. PATHWAYS INTO TEACHING IN NEW YORK CITY
Teachers in New York City can enter teaching through a number of different
pathways. 在过去, the majority of teachers have come through traditional
university-based programs, at both the graduate and undergraduate levels.
In these programs, students typically fulﬁll coursework requirements and
engage in a variety of ﬁeld experiences, including student teaching, 前
being recommended for certiﬁcation by their colleges or universities. 成为
recommended for certiﬁcation through this pathway, teachers must pass the
general knowledge exam, the Liberal Arts and Science Test (LAST), a content
specialty test (CST), and an assessment of teaching skills. They also must
complete a university-based program that is registered with the State of New
York.2 In what follows, we refer to this pathway as “college recommended.”

Teachers can also be certiﬁed through a process of individual evaluation.
Such teachers fulﬁll similar requirements to those within a traditional pro-
公克, including student teaching, but they can fulﬁll these requirements at
different institutions, and even through distance learning. Once they have
completed the requirements for certiﬁcation, they submit their transcripts to

在 1998 the New York State Board of Regents revised the registration requirements of teacher
education programs to be effective for graduates applying for certiﬁcation beginning in February
2004. These new requirements reﬂect a substantial increase in preparation requirements. 然而,
very few of the teachers employed in this analysis, all of whom were certiﬁed prior to September
2003, would have been affected by these new requirements. In subsequent analysis we will explore
the effect of these new requirements.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

179

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

the State Department of Education for review. The state then either issues a
certiﬁcate or stipulates the requirements that must be met before a certiﬁcate
can be awarded.

As the need for teachers to ﬁll the city’s classrooms intensiﬁed, an increas-
ing number of teachers began to enter teaching without prior preparation.
These teachers were hired under temporary licenses, which enabled them to
serve as the teacher of record in the classroom, but without having fulﬁlled
certiﬁcation requirements. In the spring of 2002, New York City employed
12,400 teachers with temporary licenses (New York State Department of Ed-
教育). The New York State Regents required that all teachers in New York
City be certiﬁed by September 2003—in theory ending temporary certiﬁca-
的. While the number of teachers holding temporary licenses has declined
戏剧性地, New York State has allowed the limited use of modiﬁed tempo-
rary licenses to be issued for school districts that are experiencing shortages of
certiﬁed teachers. 然而, teachers employed with these modiﬁed temporary
licenses must have completed at least twenty-seven credit hours of a prepara-
tion program and must be actively moving toward certiﬁcation. 此外,
they cannot teach in low-performing schools, and their licenses are valid only
for one year and are currently set to expire following the 2004–5 school year.
不出所料, these new policies resulted in a shortage of certiﬁed teach-
ers for New York City. Because of these and similar shortages in other school
districts, the Regents approved a framework for alternative routes into teach-
英, “designed to attract highly competent people who possess a bachelor’s
degree with a major in the subject they plan to teach, but initially lack courses
in teaching” (Ofﬁce of Teaching Initiatives, New York State Department of
教育). Participants in these alternative-route programs are expected to
complete two hundred hours of preservice training and pass the LAST and the
CST before entering the classroom. These teachers are issued “Transitional
B” certiﬁcates, good for three years, following the introductory component. 作为
teachers of record, they are expected to enroll in teacher education programs at
partner colleges to fulﬁll certiﬁcation requirements. The New York City Teach-
ing Fellows, Teach for America, and the Teaching Opportunity Program (TOP)
are all examples of alternative-route programs that provide teachers with Tran-
sitional B certiﬁcates. Participants enrolled in alternative-route programs must
fulﬁll the same requirements as all other candidates for teaching certiﬁcates;
因此, by the end of their programs, they have completed a similar set of courses
to those taken by graduates of college-recommended programs. 然而, 这
costs of entering teaching through an alternative route are substantially less
for the individual teacher than the costs of traditional university-based teacher
preparation, both because alternative-route teachers earn a salary throughout
their training and because, in the Teaching Fellows program and Teach for

180

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

美国, they pay a reduced rate for the coursework taken. 反过来, the cost
to the City is higher for these teachers because of the subsidized education.3

Teach for America recruits teachers for high-poverty schools in a num-
ber of areas around the country. They recruit teachers nationally, 瞄准
recent graduates of elite colleges and universities; the recruitment process
is extremely selective. Once corps members are selected, they must attend a
summer training institute, run by Teach for America, prior to being placed in a
classroom. The curriculum is designed by TFA and includes six components,
in addition to a summer teaching experience in New York City. Once corps
members begin teaching, they continue to take courses with a local partner
university. TFA requires a two-year commitment; at the end of this period,
corps members earn certiﬁcation, and many also earn a master’s degree.

The NYC Teaching Fellows (Fellows) began in the spring of 2000 to ad-
dress shortages in NYC public schools. In its recruitment the program targets
midcareer professionals as well as recent college graduates. The Teaching Fel-
lows program has grown from 325 in its ﬁrst year to more than 2,000 Fellows
in 2004–5. The Fellows program is one of the largest alternative-route pro-
grams in the country. 现在, 多于 6,000 Fellows are teaching in
NYC schools (New York City Teaching Fellows 2005). Prior to entering the
classroom as teachers of record, Fellows complete an introductory component,
usually offered in the summer, which includes some time in local classrooms.
The courses are taught by instructors at the partner universities. Once fel-
lows begin teaching, they continue to take classes at their partner institution.
Most Teaching Fellows complete their programs within two years. Teaching
Fellows are generally older than TFA corps members, and approximately 20
percent of Fellows have completed graduate degrees. Teaching Fellows are
typically placed in shortage subjects and schools and in the last few years
are more likely to teach math, 科学, and special education than childhood
教育.

While TFA and the Teaching Fellows are the focus of this article, 他们
are not the only alternative routes serving New York City. 例如, 这
Teaching Opportunity Program (TOP) is a collaborative initiative between the
纽约城市大学 (CUNY) and the NYC Department of Education
to produce middle and high school math, 科学, and Spanish teachers. 帕-
ticipants in TOP also take part in an intensive summer program run by a
CUNY campus that includes experiences in local schools. Once they enter

The City covers the cost of the summer training program and stipend (关于 $2,500) for Teaching Fellows. 此外, New York City pays an additional $8,000 toward the cost of their master’s
degree over the next two years, while each of the Teaching Fellows pays $4,000 toward these
degrees.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

181

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

桌子 1. Trends in the Numbers of First-Time New York City Teachers Entering through Different Pathways,
1996–2004年

Path of Entry

1996 1997 1998 1999 2000 2001 2002 2003 2004 全部

College recommended

1,260 2,192 2,901 2,840 2,618 2,375 2,018 2,434 2,192 20,830

Individual evaluation

433

740

766

736

657

534

517

712

682

5,777

Teaching Fellow
Teach for America∗

383 1,121 1,829 2,441

5,774

118

115

189

360

782

Temporary license

1,617 2,561 2,927 3,495 3,886 4,017 4,230 2,671

607 26,011

其他

Prior New York
State teaching

全部

129

172

192

268

543

912 1,106 1,263

2,896

115

102

110

128

154

235

237

1,221

3,406 5,716 6,881 7,365 7,539 8,098 9,067 9,176 7,782 65,030

∗Prior to 2001, Teach for America members were classiﬁed as temporary-license teachers.

the classroom, they continue to take courses at CUNY that count toward both
their certiﬁcation and master’s degrees. TOP participants generally complete
their requirements for certiﬁcation and an master’s degree in two to three
年, after which they are committed to teaching in NYC public schools for
an additional two years.

The distinction between alternative and traditional routes can be quite
blurry. 例如, many participants in traditional graduate programs in
teacher education apply for an internship certiﬁcate when they have com-
pleted sufﬁcient coursework. With this certiﬁcate they are able to become the
teacher of record in the classroom, earning a salary while they complete the
rest of their program and obtain a master’s degree. Given this blurry nature of
the distinction between alternative and traditional routes as well as the many
pathways into teaching in New York City, there are many ways by which to dis-
tinguish pathways. For much of our analyses we divide teachers into six groups
as deﬁned by their pathway into their ﬁrst job in New York City. These groups
是 (1) college-recommended; (2) individual-evaluation; (3) Teaching Fellows;
(4) Teach for America; (5) temporary license; 和 (6) 其他. The “other” group
includes all teachers not ﬁtting in the other ﬁve categories, including intern-
ship certiﬁcates, Transitional B teachers, and those with certiﬁcation through
reciprocity agreements with other states.

桌子 1 shows the trends in the number of ﬁrst-year New York City teachers
by pathway. The ﬁrst trend to notice is the substantial increase in new teachers
during the late 1990s and early 2000s. 在 1996, 3,406 teachers began working
in New York City for the ﬁrst time. The size of this group peaked in 2003, 什么时候
9,176 new teachers entered the school system. This increase results from an
abnormally small number of new hires in 1996, due to budget restrictions,

182

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

combined with increasing enrollments, smaller class sizes, and increasing
attrition over the period. The rise in the number of new teachers corresponded
to an increase in the number of college-recommended new teachers from
1,260 在 1996 到 2,840 在 1999, but this contributed only a portion of the
total number of new teachers. 此外, there was a large increase in the
number of teachers with temporary licenses: 从 1,617 in 1995–96 to over
4,000 in 2000–2001 and 2001–2. Prior to the 1999–2000 academic year
there were no Teaching Fellows and only a few Teach for America teachers,
but these groups grew rapidly over the next four years, 提供 2,441 和 350
teachers in 2003 和 2004, 分别. The rapid increase in the size of these
programs corresponds to a substantial decrease in the number of temporary-
license teachers and a small decrease in the number of college-recommended
教师, indicating that the majority of the alternative route teachers are ﬁlling
positions that previously had been ﬁlled by teachers with temporary licenses.
桌子 2 describes the characteristics of teachers entering New York City
public schools through different pathways. Differences are evident across
pathways, with teachers entering through alternative routes demonstrating
better performance on measures of academic preparation and/or general abil-
ity than those from other pathways. In the 2003–4 academic year, 例如,
none of the new TFA teachers and less than 2 percent of the Teaching Fellows
had failed the general knowledge certiﬁcation exam the ﬁrst time they took it,
compared with 16 percent of college-recommended and individual-evaluation
teachers and 23 percent of temporary-license teachers. 相似地, 44 百分
of Teaching Fellows and 70 percent of TFA members graduated from highly
competitive colleges (as rated by Barron’s as most competitive or highly com-
petitive), compared with much lower rates from the other pathways.

A relevant question is whether the introduction of the Teaching Fellows and
TFA programs altered the characteristics of teachers entering through other
routes. Given the economic incentives, it would not be surprising if it had. 这
trends in table 2 show some indication of this, but the effects appear small.
例如, the scores on the LAST for college-recommended teachers drops
slightly after the introduction of the Teaching Fellows program in 2000–2001.
然而, the proportion of college-recommended and individual-evaluation
teachers from more competitive colleges increases over this time, 而
proportion from the least competitive colleges decreases.

总之, the introduction of new pathways into teaching that reduced
both the tuition and the time costs of preservice preparation attracted many
new teachers. These new teachers primarily replaced temporary-license teach-
呃, a group that had been growing in the years prior to the introduction of
the new programs. The alternative-route teachers have stronger measures of
qualiﬁcations than teachers entering through any other route, 尤其是

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

183

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

桌子 2. Trends in the Characteristics of Teachers Entering New York City for the First Time by Pathway,
1996–2004年

Path of Entry

1996 1997 1998 1999 2000 2001 2002 2003 2004 全部

大学

recommended

0.136 0.144 0.143 0.132 0.117 0.126 0.134 0.145 0.162 0.137

Proportion who failed the general knowledge certiﬁcation exam

个人

评估

Teaching Fellow

Teach for America

0.149 0.14

0.172 0.169 0.173 0.15

0.165 0.172 0.161 0.162

0.056 0.106 0.065 0.018 0.055

0.037 0.02

0.011 0

0.009

Temp license

0.369 0.358 0.362 0.332 0.364 0.385 0.389 0.329 0.227 0.357

大学

recommended

252.08 250.01 249.97 250.06 250.4 248.39 247.54 247.17 246.07 248.86

Score on liberal arts and science test − standard deviation =

个人

评估

Teaching Fellow

Teach for America

262.37 254.59 252.37 250.68 249.84 249.03 249.02 246.38 247.23 249.75

266.83 256.34 261.84 267.46 263.23

273.5 271.62 270.67 275.71 273.33

Temp license

233.71 233.56 234.8 236.51 232.59 229.39 229.41 234.4 241.69 232.97

大学

recommended

0.107 0.1

0.085 0.081 0.095 0.102 0.107 0.123 0.114 0.1

Proportion from most or highly competitive colleges

个人

评估

Teaching Fellow

Teach for America

0.135 0.147 0.139 0.151 0.154 0.098 0.112 0.135 0.151 0.138

0.361 0.279 0.401 0.442 0.391

0.677 0.462 0.778 0.695 0.694

Temp license

0.14

0.144 0.14

0.136 0.125 0.115 0.126 0.142 0.175 0.132

大学

recommended

0.233 0.25

0.269 0.256 0.259 0.251 0.235 0.213 0.217 0.245

Proportion from less or least competitive colleges

个人

评估

Teaching Fellow

Teach for America

0.199 0.174 0.212 0.241 0.228 0.312 0.268 0.269 0.281 0.241

0.18

0.15

0.114 0.104 0.121

0.04

0.024 0.033 0.03

Temp license

0.262 0.251 0.244 0.27

0.27

0.282 0.272 0.247 0.229 0.264

temporarily licensed teachers they largely replaced. While the presence of the
alternative routes is likely to have drawn some teachers away from traditional
preparation because they are less costly for the student to complete, the change
has not led to a substantial drop in numbers or in the average measured edu-
cational achievement of new college-recommended teachers. 最近几年,
Teaching Fellows placements are largely targeted on difﬁcult-to-staff subjects,
such as middle and high school math and science, and special education,
subjects that attract very few traditionally prepared applicants.

184

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

3. 数据
Estimating the effect of teachers on student achievement requires good student
test score data. At present there are very few places that have data that would
allow an examination of the effects of teacher pathways into teaching on the
performance of the students they teach. New York City is one such place and,
重要的是, it provides an extraordinary range and quantity of teaching and
learning environments to examine these relationships. New York State gives
statewide student exams in mathematics and English language arts in fourth
and eighth grades. 此外, the New York City Department of Education
tests third, ﬁfth, sixth, and seventh graders in these subjects. All the exams are
aligned to the New York State learning standards; each set of tests is scaled
to reﬂect item difﬁculty, and the tests are equated across grades and over
time.4 Tests are given to all registered students with limited accommodations
and exclusions. 因此, for nearly all students the tests provide a consistent
assessment of learning for a student from one grade to the next.

Recent articles by Ballou (2002) and Kane and Staiger (2001) raise con-
cerns about how achievement tests are being used in accountability systems.
Sanders (2003) and Rogosa (2002) argue that some of these concerns are
overstated or mischaracterized or can be addressed through proper design of
accountability systems. Because the score is the dependent variable, its mea-
surement error reduces the precision of parameter estimates but does not
create bias. Our large sample sizes help compensate for potential increases
in the standard errors of the estimates. 此外, our goal is to accurately
measure not the value added by individual teachers and schools but, 相当,
the relationship between scores and teacher pathways, groups that are sub-
stantially larger than individual teachers or small schools. Statistically, 这
requires far less precision in achievement measures than that required for an
accountability system.

To conduct the analysis of the relationship between pathways into teach-
ing and student achievement, we create a student database with student exam
scores, lagged scores, and characteristics of students and their peers linked

4. There are some discrepancies and changes in the exams. The mathematics exams in all grades
are developed by CTB-McGraw Hill. New York State employs CTB-McGraw Hill for its fourth- 和
eighth-grade ELA exams. 在 2003 New York City switched from CTB to Harcourt Brace for its
third-, ﬁfth-, sixth-, and seventh-grade exams. At that time there was an equating study done to
accommodate the switch in exams producers. The format and dimensions of the State and City
exams differ in some ways, and there have been some changes in item pools over time. 最后, 这
State offers the fourth- and eighth-grade mathematics exams in the ﬁrst two weeks of May, 尽管
the City offers the third, ﬁfth-seventh mathematics exams the last week of April or the ﬁrst week of
可能. The State offers the fourth- and eighth-grade ELA exams in the ﬁrst week of February, 尽管
the City offers the third-, ﬁfth-, sixth-, and seventh-grade ELA exams in the second week of April.
Because of these discrepancies, we have normalized each exam by grade and year and included
grade and year dummies.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

185

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

to their schools, 教师, and characteristics of those teachers, 包括
indicators of the pathway into teaching. The student data, provided by the New
York City Department of Education (NYCDOE), consists of a demographic
data ﬁle and an exam data ﬁle for each year from 1998–99 through 2003–4.
The demographic ﬁles include measures of gender, 种族, language spo-
ken at home, free lunch status, special education status, number of absences,
and number of suspensions for each student who was active in any of grades
3–8 that year—approximately 450,000 到 500,000 students each year. 他们
also include fall and spring codes for the student’s school, ofﬁcial class (home-
room), 和, for upper-grade students for whom the homeroom teacher often is
not the subject matter teacher, the course and section number for the English
language arts and mathematics classes.

The exam ﬁles include, 除其他事项外, the year in which an exam was
给定, the grade level of the exam (例如, a grade between third and eighth), 和
the student’s scaled score on the exam. For most years, the ﬁles contain scores
for approximately 65,000 到 80,000 students in each grade. The only signif-
icant exception is that the ﬁles contain no scores for seventh-grade English
language arts in 2002 because the New York City Department of Education
is not conﬁdent that exam scores for that year and grade were measured in a
manner that was comparable to the seventh-grade English language arts exam
in other years.

Using these data, we construct a set of records with a student’s current
exam score and his or her lagged exam score. For this purpose, a student
was considered to have value-added information in cases where we had a
score in a given subject (ELA or math) for the current year and a score for the
same subject in the immediately preceding year for the immediately preceding
年级. We do not include cases in which a student took a test for the same
grade two years in a row, or where a student skipped a grade.

While NYCDOE does not maintain an identiﬁer linking students directly to
their teachers, in most cases we were able to create our own links using school
and course identiﬁers because the NYCDOE’s data systems track the courses
taken by each student and the courses taught by each teacher. Based on advice
from NYCDOE staff, we matched students in grades 3–5 to teachers based
on the homeroom identiﬁer, and matched students in grades 6–8 based on
the section of a course being taught. 很遗憾, some middle schools do
not participate in NYCDOE’s middle school performance assessment system
(MSPA), and in those cases the course-section identiﬁer is not linked centrally
to teachers. Because of this, our matches are much lower for grades 6–8
than for grades 3–5. The matches for grade 6 in particular are lower than for
other grades. 平均而言, in grades 3–5 we have valid teacher identiﬁcation
for approximately 96 到 98 percent of students with exam scores; for grades 7

186

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

和 8 we have valid matches for about two-thirds of students with exam scores;
and for grade 6 we have valid identiﬁcations for about half of the students.
全面的, we had valid matches for approximately 80 percent of students with
exam scores.5

To enrich our data on teachers, we match New York City teachers to data
from New York State Education Department databases, using a crosswalk
ﬁle provided by NYCDOE that links their teacher ﬁle reference numbers to
unique identiﬁers compatible with both databases. We draw variables for NYC
teachers from New York State data ﬁles as follows:

(西德:1) Teacher Experience: For teacher experience, we use transaction-level data
from the NYCDOE Division of Human Resources to identify when indi-
viduals joined the NYCDOE payroll system in a teaching position. 什么时候
this information is missing or when the value is less than the value in the
NYSED personnel master ﬁles, we use the NYSED data.

(西德:1) Teacher Demographics: We draw gender, 种族, and age from a com-
bined analysis of all available data ﬁles, to choose most-common values for
个人.

(西德:1) Undergraduate: We identify the institutions from which individual teach-
ers earned their undergraduate degrees from the NYS Teacher Certiﬁcation
Database (TCERT) and combine it with the Barron’s ranking of college se-
lectivity to construct variables measuring the selectivity of the college from
which each teacher graduated.

(西德:1) Test Performance: We draw information regarding the teacher certiﬁcation
exam scores of individual teachers and whether they passed on their ﬁrst
attempts from the NYS Teacher Certiﬁcation Exam History File (EHF).
(西德:1) Pathway: Initial pathway into teaching comes from an analysis of teacher
certiﬁcation applications plus separate data ﬁles for individuals who
participated in Teach for America, the Teaching Fellows Program, 和
the New York City Teaching Opportunity Program.

The average attributes of sixth- through eighth-grade students who are matched to teachers com-
pared to those who are not matched are substantially the same with a few exceptions. Matches in
sixth and seventh grades had lower achievement gains than nonmatches (.019 与. .105 和 .055 与.
.061 分别). Sixth-grade achievement scores for nonmatches also exceeded those for matches
(.212 与. .159). 此外, matches were more likely to include black and Asian students and less
likely to include Hispanics than nonmatches. For the remaining student variables (seventh- 和
eighth-grade achievement scores, absences, suspensions, white students, home language, 性别,
free lunch status) differences were frequently statistically signiﬁcant due to the very large number
of students but did not differ by more than 5 百分. As one way of assessing the importance of
these differences, we estimated pathway effects by grade and found only small differences across
grades 6–8. We have no reason to believe that differences in achievement gains are correlated with
pathways and thus believe our pathway estimates are unbiased.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

187

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

(西德:1) College-recommended: We obtain indicators for whether an individual
had completed a college-recommended teacher preparation program, 和,
如果是这样, the level of degree obtained (bachelor’s or master’s), from NYSED’s
program completers data ﬁles.

Using these data, we construct our indicator of the pathway into teaching as fol-
lows. Any individual who is separately identiﬁed as participating in Teach for
美国, Teaching Fellows, or the Teaching Opportunity Program is coded as
entering teaching through that pathway, as appropriate. For other individuals,
we examine certiﬁcation application records to determine the earliest pathway
for which they had approval from NYSED prior to their ﬁrst teaching job in
the New York State public schools, with those pathways deﬁned as: (1) tradi-
tional college recommended; (2) individual evaluation; (3) temporary license
(individuals who failed to complete one or more requirements for a teaching
certiﬁcate but were allowed to teach under the temporary-license provisions,
whereby a school district can request NYSED to allow a speciﬁc individual to
teach in a speciﬁc school for a temporary period); (4) other certiﬁcates, 在-
cluding internship certiﬁcates, other Transitional B teachers, and those with
certiﬁcation through reciprocity agreements with other states.

有 1,035,949 student observations for the assessment of mathemat-
ics achievement and 926,958 student observations for the assessment of ELA
achievement. These observations include information on current and prior test
performance as well as information on all other variables used in the models
except free lunch program participation. There is substantial missing data on
this variable, and we run our analysis both with the reduced sample and with
this variable recoded so that those with missing information are assumed not
to be eligible.

The percent of black and Hispanic student observations are about equal at
35 到 37 百分, while just over 11 percent of student observations are Asian.6
Approximately three-quarters of the sample is eligible for free lunch, and ap-
近似地 40 percent speak a language other than English at home. As noted
多于, we have better student coverage in the elementary grades: 例如,
in math, 27 percent of the sample comes from the fourth grade, compared
和 14 percent from the sixth grade and 16 percent from the eighth grade.
正如预期的那样, the largest group of students is taught by college-recommended
教师 (38 百分), though a large proportion is taught by temporary-license

6. A table that documents the following description of the attributes of students in our sample is

available on the Education Finance and Policy Web site.

188

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

教师 (37 百分), with smaller shares for independent-evaluation teachers
(14 百分), Teaching Fellows (4 百分), and TFA members (1 百分). 这
ELA sample has a higher percent of college-recommended teachers than does
the math sample because these teachers are less likely to teach middle school
math. They are replaced there largely by temporary-license teachers. 最后,
because of issues with the tests discussed above, we standardized each test
score for each grade and year to have a standard deviation of one and a mean
of zero. The average gain in these standardized scores is zero, but the standard
deviation is 0.66 for math and 0.71 for ELA.7

Teachers differ by pathway in the characteristics of their students.8 When
we examine only ﬁrst-year teachers who began teaching in New York City
between 1998–99 and 2003–4, we ﬁnd that Teaching Fellows and TFA
teachers work with students who are more likely to be nonwhite, 贫穷的, 和
low-performing than teachers from other pathways do. 尽管, 一般,
9 percent of students are white, 少于 1 percent of students taught by TFA
teachers are white, 并且只有 3 percent of students taught by Fellows are white,
compared with 16 percent of the students of college-recommended teachers.
相似地, 大约 82 percent of the students of ﬁrst-year teachers are
eligible for free lunch; the corresponding numbers are 76 percent for college-
recommended teachers, 88 percent for Teaching Fellows, 92 percent for TFA
members, 和 83 percent for temporary-license teachers. Prior test scores show
the same pattern. They are −0.26 on average for ﬁrst-year teachers; −0.14 for
college-recommended; −0.29 for temporary-license; −0.38 for Fellows; 和
−0.51 for TFA members.9

TFA teachers are the youngest; college-recommended teachers, the next
youngest; and the other pathways include teachers of similar average ages.
Most teachers are female, particularly college-recommended teachers. Fellows
have the highest percent of male teachers, 在 33 百分. Most teachers are
白色的. Temporary-license teachers are the most racially diverse, 和 19 百分
black and 28 percent Hispanic. We ﬁnd that college-recommended teachers
are the least likely to have attended a highly competitive college, while TFA
members are by far the most likely. The temporary-license teachers performed
the least well on the general knowledge certiﬁcation exam, 与超过 30 百分
failing on the ﬁrst try, while no TFA member in this sample failed.

The standard deviation of the gain is similar across grade levels.

7.
8. A table showing the details of attributes of teachers and students by the pathways is available on the

Education Finance and Policy Web site.

9. The greater proportion of Hispanic students relative to black students taught by TFA members is

largely due to the placement of TFA teachers in Washington Heights and the South Bronx.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

189

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

4. 方法
How do the achievement gains of students differ by the teaching pathway of
their teachers? Speciﬁcally, do students of teachers who enter the classroom
with reduced coursework preparation and few prior ﬁeld experiences achieve
gains that are higher or lower than they would have if they were taught by tradi-
tionally certiﬁed or temporary-license teachers? Analyses considering whether
teachers from one pathway are more effective than those from another path-
way need to account for differences both in the students that teachers teach
and in the work environments in which they teach. Different pathways into
teaching lead teachers into schools with different characteristics and differ-
ent students. 因此, we must account for these differences in the matching
of teachers to schools and students if we are to accurately assess the effect
of pathway. Because TFA members, 例如, are assigned to schools that
have traditionally been difﬁcult to staff, the achievement of their students may
be lower than the achievement of other students; but we would not want to
attribute this difference to a negative causal relationship between TFA and stu-
dent achievement. 重要的, students may be similar in many measured
characteristics such as race/ethnicity or free lunch participation, but have dif-
ferent achievement trajectories due, 例如, to the school they attend.
If teachers from different pathways are differentially assigned to schools in
which students are likely to learn more (或更少), then we need to account for
these differences.

While there is much agreement among researchers about the need to
account for these assignment and selection issues, there is less agreement
about which model speciﬁcation is the most accurate for this accounting.
Because of this, we choose to run a number of different speciﬁcations in order
to test the robustness of the estimated effects. The base model that we used is
summarized by equation 1.

Aisgty = γ

+ C

1Ais’g(g−1)t’(y−1) + γ2Siy + C j t γ3 + Ejtγ4 + Pjγ5 + π s

0
+ π g + π y + (西德:3)isgty

(1)

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Here the standardized achievement level (test score) A of student i in school
s in grade g with teacher t in year y is a linear function of the student’s test
score in the prior year plus the square of this, characteristics of the student S,
characteristics of the other students in the same grade with the same teacher in
that year C, the teacher’s teaching experience E , and the pathway the teacher
took into teaching P. Student characteristics include gender, race/ethnicity,
poverty status, days absent during the prior year, and suspensions in the prior
年. The aggregate (teacher by grade by school by year) student characteristics

190

EDUCATION FINANCE AND POLICY

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

include race/ethnicity, poverty status, average attendance in the prior year,
average suspensions in the prior year, average student test scores in the prior
年, and the standard deviation of student test scores in the prior year. Teach-
ing experience is measured by dummy variables for each year of teaching from
the ﬁrst year through the twentieth and then an additional dummy variable
for experience greater than twenty years. 此外, the model includes ﬁxed
effects for years, grades, and schools. The standard errors are clustered at the
teacher level to account for the fact that teacher pathway is a teacher-level
variable.

This model controls for all the attributes of students that typically re-
main constant from one year to the next, such as parental support and home
环境. It also controls for all the characteristics of schools that do not
change over the 2000–2004 period by including a school ﬁxed effect. 一
implication of this approach is that the effectiveness of a teacher pathway is
compared only to other pathways in the same school. 这样, 我们不
attribute to teachers or their pathways any effect that may differ from one
school to another but that is not measured by the included school or class
characteristics.10 This is generally a conservative approach to the identiﬁcation
of the effects of pathway effectiveness. We ﬁnd in our speciﬁcation checks
that the results are robust to an even more conservative approach of includ-
ing school-by-year-by-grade ﬁxed effects that effectively compare a teacher’s
pathway to other pathways in the same school, 年, and grade. 最后, 我们
assume that included student and class variables control for any within-school
assignment of teacher pathways to classes that may be correlated with student
achievement.

This model assumes that the returns to experience are similar across path-
方法. This may not be the case. Teachers from different pathways enter with
very different experiences and strengths. We might imagine that teachers who
have had few classroom experiences prior to teaching may gain differentially
in their teaching effectiveness over the ﬁrst few years. This may be especially
true for teachers entering through alternative certiﬁcation pathways. They may

10. School ﬁxed effects account for unobservable differences across schools. We cannot use the same
approach for classrooms because there is only one teacher in each classroom. Because of this, 我们用
controls for classroom characteristics; 然而, there may still be omitted unobserved characteristics
related to teacher pathways. In order to assess the likelihood of omitted variables bias, we examine
whether within schools teachers entering from different pathways systematically teach students with
different observed characteristics. If classrooms differ in observables by pathway, they probably
are more likely to differ in unobservables as well. We ﬁnd that there is no statistically signiﬁcant
difference in lagged student absences by pathway and that the only difference in lagged achievement
scores are between temporary-license teachers and most other teachers. Teachers with temporary
licenses teach students who enter the class with lower test scores. 因此, while it is possible that our
estimates may be affected by omitted variable bias, we believe this is unlikely.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

191

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

gain more because they are learning what many teachers learned prior to start-
英, either through on-the-job-training or through their ongoing education in
master’s programs, plus the typical things that teachers learn over the ﬁrst
couple of years. 或者, they may be so overwhelmed by ﬁguring out
how to get the class in order or they may have such little training in areas such
as child development or the teaching of speciﬁc subject matters that they have
less time and less of a framework to help them grow as teachers. In a second
模型, we add interactions between pathway and teacher experience in order
to estimate these differences in returns to experience.

5. 结果
Tables 3a and 3b give the results for the base model for mathematics achieve-
ment and English language arts achievement, 分别. The base models
aggregate the effects of students in grades 4 通过 8. Two sets of estimates
are presented: the total pathway effects that summarize the average results by
路径; and a set of results that indicate the differential effects of pathway
by the level of teacher experience. First consider the student-level variables.
正如预期的那样, there is a strong positive relationship between the current score
and the lagged score. Black and Hispanic students, along with students receiv-
ing free lunch, demonstrate smaller gains over the year, in both mathematics
and ELA, than white students and students from higher-income families. 这
estimate for black students is particularly large. The standard deviation in
test score gain over the year for mathematics is approximately 0.66 点,
indicating that black students gain approximately 16.5 percent of a standard
deviation less than white students over the course of a year, holding all else
持续的. Female students gain less in math and more in ELA than do male
学生. Aggregate student measures also appear to affect test score gains. 在
特别的, classes with a greater proportion of black or Hispanic students or a
lower average starting test score gain substantially less on average than other
classrooms.

Other studies have found (Rivkin, Hanushek, and Kain 2005; Rockoff
2004) that teacher experience is an important inﬂuence on student achieve-
蒙特. Tables 3a and 3b provide estimates of these effects for teachers by
their level of experience. The experience estimates for this model include two
components—the effect of additional years of teaching on the effectiveness
of individual teachers and the differential selection of teachers across years of
experience both as a result of quitting and through the change in the teacher
labor market over time. 例如, if many of the least able teachers left
after their ﬁrst year, then student achievement would increase for second-year
teachers relative to ﬁrst-year teachers even if the teachers who remained in

192

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

D. Boyd, 磷. Grossman, H. Lankford, S. Loeb, 和 J. Wyckoff

the second year had not improved. Estimates that hold constant the selection
of teachers provide insight on the extent to which average student achieve-
ment increases as teachers gain the knowledge that comes with experience. 如果,
然而, the question is taken from the student perspective—to what extent
are students advantaged by having a second- or third-year teacher relative to
a novice?—then it is important to include both the individual and selection
effects of experience. Since our interest is in understanding the differences
across pathways, which potentially include both differential gains in individual
teacher effectiveness as well as differential retention, we focus primarily on
the total effect.

Tables 3a and 3b show differences in the student achievement gains asso-
ciated with teaching experience. 例如, for math achievement, 第二-
year teachers have average gains that are about 0.05 点 (7.6 percent of a
标准差) larger than those of ﬁrst-year teachers. Third-year teachers
have students that gain approximately .02 points more students of second-year
教师 (.067 - .050 or about 3 percent of a standard deviation difference
in test score gains). There are few gains to experience after the third year of
教学.

The variables listed under Teacher Pathway of tables 3a and 3b provide the
estimates of the effects of pathways into teaching. These effects are relative to
college-recommended teachers. For the total pathway results for mathemat-
集成电路, we see that the coefﬁcient on Individual Evaluation is −.012, indicating
that students with teachers who entered New York City through this pathway
show a test score gain lower than 1.8 percent of a standard deviation (这
standard deviation of the gain is 0.66) than that of students with college-
recommended teachers. Students of TFA members have similar test scores to
those of college-recommended teachers, while students of Teaching Fellows,
temporary-license teachers, and teachers who entered through other pathways
show the smallest gains, 大约 3.3 percent of standard deviation lower
than college-recommended. The relationships for ELA are somewhat different.
Individual-evaluation teachers have students with gains about equal to those
of college-recommended teachers, while students of Teaching Fellows and
TFA teachers show lower gains. Temporary-license teachers and other teach-
ers demonstrate effects that are somewhat lower than college-recommended
teachers and higher than Teaching Fellows.

The effect of experience by pathway is shown in the Experience by Pathway
column and is summarized in table 4. When we examine the results for
pathway by experience, teachers improve with experience over the ﬁrst few
年. 然而, there appears, in this speciﬁcation, to be little difference in
returns to experience across pathways. While the point estimates suggest that
Fellows and TFA teachers gain relative to college-recommended teachers as

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

193

1
9
4

我

乙
D
U
C
A
时间
氧
氮
F
我
氮
A
氮
C
乙

A
氮
D
磷
氧
L
我
C
是

Table 3a. Base Model Student Achievement for Math with School Fixed Effects

STUDENT MEASURES

TEACHER EXPERIENCE

TEACHER PATHWAY

全部的
Pathway

经验
by Path

全部的
Pathway

经验
by Path

Lag score

Lag score sqrd

女性

Other ethnicity

Asian

Hispanic

黑色的

English home

Free lunch

Lagged absent

Lag suspended

0.659∗∗∗
(0.002)

0.001
(0.001)

−0.009∗∗∗
(0.001)

−0.088∗∗∗
(0.011)

0.114∗∗∗
(0.003)

−0.083∗∗∗
(0.003)

−0.109∗∗∗
(0.003)

−0.047∗∗∗
(0.002)

−0.046∗∗∗
(0.002)

−0.004∗∗∗
(0.000)

−0.052∗∗∗
(0.004)

0.659∗∗∗
(0.002)

0.001
(0.001)

−0.009∗∗∗
(0.001)

−0.088∗∗∗
(0.011)

0.114∗∗∗
(0.003)

−0.083∗∗∗
(0.003)

−0.109∗∗∗
(0.003)

−0.047∗∗∗
(0.002)

−0.046∗∗∗
(0.002)

−0.004∗∗∗
(0.000)

−0.052∗∗∗
(0.004)

2 年

3 年

4 年

5 年

6 年

7 年

8 年

9 年

10 年

11 年

12 年

0.048∗∗∗
(0.005)

0.067∗∗∗
(0.005)

0.073∗∗∗
(0.006)

0.077∗∗∗
(0.006)

0.080∗∗∗
(0.007)

0.068∗∗∗
(0.007)

0.075∗∗∗
(0.008)

0.070∗∗∗
(0.008)

0.074∗∗∗
(0.008)

0.071∗∗∗
(0.008)

0.065∗∗∗
(0.009)

0.050∗∗∗
(0.008)

0.067∗∗∗
(0.009)

0.079∗∗∗
(0.008)

0.083∗∗∗
(0.009)

0.085∗∗∗
(0.009)

0.073∗∗∗
(0.009)

0.080∗∗∗
(0.010)

0.075∗∗∗
(0.010)

0.079∗∗∗
(0.010)

0.076∗∗∗
(0.010)

0.070∗∗∗
(0.011)

Fellows

TFA

Temporary

其他

IE∗Exp2

IE∗ Exp3

IE∗Exp>3

Fellows∗Exp2

Fellow*Exp3

Fellows*Exp>3

全部的
Pathway

−0.012∗∗
(0.005)

−0.023∗∗∗
(0.008)

0.007
(0.015)

−0.021∗∗∗
(0.004)

−0.021∗∗∗
(0.008)

经验
by Path

−0.014
(0.017)

−0.020∗
(0.011)

0.009
(0.018)

−0.011
(0.009)

−0.023∗
(0.013)

−0.014
(0.023)

−0.005
(0.024)

0.002
(0.018)

0.016
(0.016)

0.02
(0.023)

−0.105∗∗∗
(0.025)

H
氧
瓦

C
H
A
氮
G
乙
S

我

氮

乙
氮
时间
右
是

右
乙
问
U

我

右
乙
中号
乙
氮
时间
S

A
L
时间
乙
右

时间
H
乙

时间
乙
A
C
H
乙
右

瓦
氧
右
K
F
氧
右
C
乙

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Class Average Measures

Other ethnicity

Hispanic

黑色的

Asian

English home

Free lunch

Lagged absent

Lag suspended

Lag score

Std. dev. score deviat.

持续的

观察结果
Rt-squared

−0.177
(0.115)

−0.182∗∗∗
(0.028)

−0.230∗∗∗
(0.026)

−0.061∗
(0.033)

0.042∗∗
(0.019)

−0.057∗∗∗
(0.013)

−0.010∗∗∗
(0.001)

−0.078∗∗
(0.033)

0.154∗∗∗
(0.005)

0.006
(0.008)

其他

0.321∗∗∗
(0.029)

960,970
0.64

−0.18

(0.115)

−0.183∗∗∗
(0.028)

−0.231∗∗∗
(0.026)

−0.061∗
(0.033)

0.042∗∗
(0.019)

−0.057∗∗∗
(0.013)

−0.010∗∗∗
(0.001)

−0.077∗∗
(0.033)

0.153∗∗∗
(0.005)

0.006
(0.008)

0.324∗∗∗
(0.029)

960,970
0.64

13 年

14 年

15 年

16 年

17 年

18 年

19 年

20 年

> 20 年

2000

2001

2002

2003

0.062∗∗∗
(0.009)

0.080∗∗∗
(0.010)

0.056∗∗∗
(0.009)

0.054∗∗∗
(0.009)

0.053∗∗∗
(0.010)

0.059∗∗∗
(0.011)

0.040∗∗∗
(0.012)

0.049∗∗∗
(0.012)

0.055∗∗∗
(0.007)

年

0.018∗∗∗
(0.004)

0.008∗
(0.004)

0.011∗∗∗
(0.004)

0.004
(0.004)

0.067∗∗∗
(0.011)

0.085∗∗∗
(0.011)

0.061∗∗∗
(0.011)

0.059∗∗∗
(0.011)

0.058∗∗∗
(0.012)

0.064∗∗∗
(0.013)

0.045∗∗∗
(0.014)

0.054∗∗∗
(0.014)

0.059∗∗∗
(0.009)

0.018∗∗∗
(0.004)

0.008∗
(0.004)

0.012∗∗∗
(0.004)

0.004
(0.004)

TFA*Exp2

TFA*Exp3

TFA*Exp>3

Temp*Exp2

Temp*Exp3

Temp*Exp>3

Other*Exp2

Other*Exp3

Other*Exp>3

−0.004
(0.025)

0.028
(0.045)

−0.012
(0.045)

−0.011
(0.011)

−0.005
(0.012)

−0.012
(0.010)

0.02
(0.020)

0.021
(0.022)

−0.003
(0.016)

0.007
(0.005)

0.158∗∗∗
(0.011)

0.185∗∗∗
(0.011)

0.197∗∗∗
(0.011)

Grade

0.006
(0.005)

0.158∗∗∗
(0.011)

0.186∗∗∗
(0.011)

0.198∗∗∗
(0.011)

Notes: Robust standard errors in parentheses. ∗signiﬁcant at 10%; ∗∗signiﬁcant at 5%; ∗∗∗ signiﬁcant at 1%. Results for TFA in the third year should be interpreted
with caution due to small sample sizes.

乙
哦
y
d
,

磷.

r
哦
s
s
米
A
n
,

L
A
n
k
F
哦
r
d
,

L
哦
e
乙
,

A
n
d

J
.

瓦
y
C
k
哦
F
F

1
9
5

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

1
9
6

我

乙
D
U
C
A
时间
氧
氮
F
我
氮
A
氮
C
乙

A
氮
D
磷
氧
L
我
C
是

Table 3b. Base Model Student Achievement for ELA with School Fixed Effects

STUDENT MEASURES

TEACHER EXPERIENCE

TEACHER PATHWAY

全部的
Pathway

经验
by Path

全部的
Pathway

经验
by Path

Lag score

Lag score sqrd

女性

Other ethnicity

Asian

Hispanic

黑色的

English home

Free lunch

Lagged absent

Lag suspended

0.606∗∗∗
(0.002)

−0.004∗∗∗
(0.001)

0.058∗∗∗
(0.002)

−0.100∗∗∗
(0.012)

0.051∗∗∗
(0.004)

−0.088∗∗∗
(0.003)

−0.115∗∗∗
(0.003)

−0.019∗∗∗
(0.002)

−0.069∗∗∗
(0.002)

−0.003∗∗∗
(0.000)

−0.057∗∗∗
(0.005)

0.606∗∗∗
(0.002)

−0.004∗∗∗
(0.001)

0.058∗∗∗
(0.002)

−0.100∗∗∗
(0.012)

0.051∗∗∗
(0.004)

−0.088∗∗∗
(0.003)

−0.115∗∗∗
(0.003)

−0.019∗∗∗
(0.002)

−0.069∗∗∗
(0.002)

−0.003∗∗∗
(0.000)

−0.057∗∗∗
(0.005)

2 年

3 年

4 年

5 年

6 年

7 年

8 年

9 年

10 年

11 年

12 年

0.030∗∗∗
(0.005)

0.042∗∗∗
(0.005)

0.041∗∗∗
(0.006)

0.052∗∗∗
(0.006)

0.056∗∗∗
(0.007)

0.060∗∗∗
(0.008)

0.054∗∗∗
(0.009)

0.058∗∗∗
(0.009)

0.041∗∗∗
(0.009)

0.033∗∗∗
(0.008)

0.041∗∗∗
(0.008)

0.047∗∗∗
(0.007)

0.058∗∗∗
(0.008)

0.062∗∗∗
(0.008)

0.063∗∗∗
(0.008)

0.067∗∗∗
(0.009)

0.066∗∗∗
(0.010)

0.060∗∗∗
(0.010)

0.065∗∗∗
(0.010)

0.048∗∗∗
(0.010)

Fellows

TFA

Temporary

其他

IE*Exp2

IE*Exp3

IE*Exp>3

Fellow*Exp2

Fellow∗Exp3

Fellows∗Exp>3

全部的
Pathway

−0.005
(0.005)

−0.030∗∗∗
(0.007)

−0.031∗∗∗
(0.012)

−0.012∗∗∗
(0.004)

−0.021∗∗∗
(0.007)

经验
by Path

0.012
(0.014)

−0.026∗∗∗
(0.010)

−0.032∗∗
(0.015)

−0.004
(0.009)

−0.013
(0.013)

−0.014
(0.021)

−0.004
(0.024)

−0.02
(0.015)

−0.011
(0.014)

0.051∗∗
(0.025)

−0.038∗
(0.021)

H
氧
瓦

C
H
A
氮
G
乙
S

我

氮

乙
氮
时间
右
是

右
乙
问
U

我

右
乙
中号
乙
氮
时间
S

A
L
时间
乙
右

时间
H
乙

时间
乙
A
C
H
乙
右

瓦
氧
右
K
F
氧
右
C
乙

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Class Average Measures

Other ethnicity

Hispanic

黑色的

Asian

English home

Free lunch

Lagged absent

Lag suspended

Lag score

Std. dev. 分数

持续的

观察结果
R−squared

−0.280∗∗∗
(0.102)

−0.182∗∗∗
(0.026)

−0.236∗∗∗
(0.027)

0.047
(0.031)

0.023
(0.016)

−0.075∗∗∗
(0.013)

−0.008∗∗∗
(0.001)

0.001
(0.031)

0.199∗∗∗
(0.004)

0.020∗∗∗
(0.007)

其他

0.315∗∗∗
(0.027)

861,698
0.59

−0.280∗∗∗
(0.102)

−0.182∗∗∗
(0.026)

−0.237∗∗∗
(0.027)

0.047
(0.031)

0.023
(0.016)

−0.075∗∗∗
(0.013)

−0.008∗∗∗
(0.001)

0.001
(0.031)

0.199∗∗∗
(0.004)

0.020∗∗∗
(0.007)

0.310∗∗∗
(0.028)

861,698
0.59

13 年

14 年

15 年

16 年

17 年

18 年

19 年

20 年

> 20 年

2000

2001

2002

2003

0.060∗∗∗
(0.009)

0.051∗∗∗
(0.009)

0.059∗∗∗
(0.009)

0.073∗∗∗
(0.012)

0.056∗∗∗
(0.012)

0.054∗∗∗
(0.012)

0.029∗∗
(0.013)

0.031∗∗
(0.012)

0.046∗∗∗
(0.007)

年

0.007∗
(0.004)

−0.004
(0.004)

0.002
(0.004)

−0.003
(0.004)

0.067∗∗∗
(0.011)

0.058∗∗∗
(0.011)

0.065∗∗∗
(0.011)

0.079∗∗∗
(0.013)

0.062∗∗∗
(0.013)

0.060∗∗∗
(0.013)

0.035∗∗
(0.014)

0.037∗∗∗
(0.013)

0.052∗∗∗
(0.008)

0.008∗
(0.004)

−0.004
(0.004)

0.003
(0.004)

−0.003
(0.004)

TFA∗Exp2

TFA∗Exp3

TFA∗Exp>3

Temp∗Exp2

Temp∗Exp3

Temp∗Exp>3

Other∗Exp2

Other∗Exp3

Other∗Exp>3

0.009
(0.022)

0.008
(0.042)

0.032
(0.044)

−0.004
(0.011)

−0.002
(0.012)

−0.011
(0.010)

0.001
(0.019)

−0.009
(0.023)

−0.011
(0.017)

0.003
(0.004)

0.124∗∗∗
(0.011)

0.146∗∗∗
(0.010)

0.156∗∗∗
(0.011)

Grade

0.003
(0.004)

0.125∗∗∗
(0.004)

0.146∗∗∗
(0.010)

0.157∗∗∗
(0.011)

Notes: Robust standard errors in parentheses. ∗signiﬁcant at 10%; ∗∗signiﬁcant at 5%; ∗∗∗signiﬁcant at 1%. Results for TFA in the third year should be interpreted
with caution due to small sample sizes.

乙
哦
y
d
,

磷.

r
哦
s
s
米
A
n
,

L
A
n
k
F
哦
r
d
,

L
哦
e
乙
,

A
n
d

J
.

瓦
y
C
k
哦
F
F

1
9
7

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE

桌子 4. Effects of Teacher’s Pathway and Years of Experience on Student
Achievement by Grade

Pathway

1 年

2 Years

3 Years

College recommended

Individual evaluation

NYC Teaching Fellows

TFA

Temporary license

其他

College recommended

Individual evaluation

NYC Teaching Fellows

TFA

Temporary license

其他

Math grades 4–8

.048

.020

.044

.053

.026

.045

ELA grades 4–8

.030

.028

-.007

.007

.022

.018

.067

.048

.067

.104

.051

.065

.042

.050

.067∗

.018

.036

.020

-.014

-.020∗

.009

-.011

-.023∗

.012

-.026∗

-.032∗

-.004

-.013

Notes: Based on estimates in table 3. Results for TFA in the third year
should be interpreted with caution due to small sample sizes.

experience increases, in general these gains are not statistically signiﬁcant in
this speciﬁcation.11

The ﬁrst six columns of tables 5a and 5b report the results of the speciﬁca-
tion checks described above.12 The consistency of the estimates across models
is striking. 例如, for math, the effect of Teaching Fellows relative to
college-recommended teachers is −.023 in the base model and −.025 in both
the model with only teachers with three or fewer years of experience and the
two-stage full model; the magnitude of the coefﬁcient increases to .031 在里面

12.

11. The estimates of the differential return to experience in the third year and beyond for TFA members
and beyond the third year for Fellows may not be accurate due to small sample sizes for these groups.
Because of this we restrict our discussion of results to groups with at least twenty-ﬁve teachers. 我们
also omit from the summary tables to be discussed below the estimates of differential returns to
experience for the teachers with greater than three years of experience, though the variables are
included in the models.
In addition to the general models presented in tables 5a and 5b, estimates that show the pathway-
experience interactions across each of these speciﬁcation is available on the Education Finance and
Policy Web site. And in addition to the speciﬁcation checks in tables 5a and 5b, we have estimated
models that attempt to control for the differential timing of the student exams between grades and
across years. 例如, the fourth-grade math exam is often given at the beginning of February.
Thus the ﬁfth-grade math exam, often given at the beginning of April, includes several months of
learning partly attributable to the fourth-grade teacher. We have estimated models that allocate each
student’s ﬁfth-grade scores proportionately to the fourth- and ﬁfth-grade teachers. The resulting
estimates are very similar to those presented in this article. Positive and negative effects are slightly
larger, and in some cases the precision of the estimates improves slightly.

198

EDUCATION FINANCE AND POLICY

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
e
d
p
A
r
t
我
C
e
–
p
d

我

F
/

1
2
1
7
6
1
6
8
8
9
6
9
e
d
p
2
0
0
6
1
2
1
7
6
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Table 5a. Speciﬁcation Checks on the Base Model for Math Achievement

Base Model

<= 3 Years Experience Student Fixed Effect on Gains School/ Year/Grade Fixed Effects IE Fellows TFA Temp Other −0.012∗∗ (0.005) −0.023∗∗∗ (0.008) 0.007 (0.015) −0.021∗∗∗ (0.004) −0.021∗∗∗ (0.008) −0.028∗∗ (0.011) −0.025∗∗∗ (0.009) −0.001 (0.016) −0.026∗∗∗ (0.006) −0.025∗∗∗ (0.010) −0.009 (0.006) −0.014 (0.009) 0.015 (0.017) −0.010∗∗ (0.004) −0.018∗∗ (0.008) −0.011∗∗ (0.005) −0.030∗∗∗ (0.007) −0.005 (0.015) −0.017∗∗∗ (0.004) −0.015∗∗ (0.007) Two Stage −0.015∗∗ (0.006) −0.025∗∗∗ (0.009) −0.004 (0.016) −0.023∗∗∗ (0.005) −0.026∗∗∗ (0.009) 2 Stage Exp <=3 −0.032∗∗ (0.013) −0.031∗∗∗ (0.010) −0.011 (0.018) −0.034∗∗∗ (0.007) −0.039∗∗∗ (0.012) No Class Controls −0.014∗∗ (0.006) −0.025∗∗∗ (0.008) 0.006 (0.017) −0.034∗∗∗ (0.005) −0.027∗∗∗ (0.008) No School Fixed Effect No School or Class −0.013∗∗ (0.006) −0.015∗ (0.008) 0.002 (0.016) −0.022∗∗∗ (0.004) −0.021∗∗∗ (0.008) −0.018∗∗∗ (0.007) −0.048∗∗∗ (0.009) −0.036∗∗ (0.018) −0.049∗∗∗ (0.005) −0.036∗∗∗ (0.009) N 960,970 320,503 960,970 960,970 960,970 320,503 960,970 960,970 960,970 Notes: Models include all variables in Table 3. Robust standard errors in parentheses. ∗signiﬁcant at 10%; ∗∗signiﬁcant at 5%; ∗∗∗signiﬁcant at 1%. Results for TFA in the third year should be interpreted with caution due to small sample sizes. D . B o y d , P. G r o s s m a n , H . L a n k f o r d , S . L o e b , a n d J . W y c k o f f 1 9 9 l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 2 0 0 I E D U C A T O N F I N A N C E A N D P O L I C Y Table 5b. Speciﬁcation Checks on the Base Model for ELA Achievement Gain as Outcome <= 3 Years Experience Student Fixed Effect on Gains School/ Year/Grade Fixed Effects IE Fellows TFA Temp Other −0.005 (0.005) −0.030∗∗∗ (0.007) −0.031∗∗∗ (0.012) −0.012∗∗∗ (0.004) −0.021∗∗∗ (0.007) 0.00 (0.010) −0.022∗∗∗ (0.008) −0.030∗∗ (0.013) −0.011∗∗ (0.006) −0.016∗ (0.009) 0.002 (0.005) −0.016∗∗ (0.007) −0.031∗∗ (0.014) −0.002 (0.004) −0.008 (0.008) −0.003 (0.004) −0.031∗∗∗ (0.006) −0.022∗ (0.012) −0.009∗∗∗ (0.003) −0.21∗∗∗ (0.006) Two Stage −0.003 (0.005) −0.032∗∗∗ (0.007) −0.033∗∗ (0.013) −0.012∗∗∗ (0.004) −0.023∗∗∗ (0.008) 2 Stage Exp <=3 −0.004 (0.010) −0.028∗∗∗ (0.009) −0.035∗∗ (0.014) −0.014∗∗ (0.006) −0.25∗∗ (0.011) No Class Controls No School Fixed Effect No School or Class −0.010∗ (0.006) −0.036∗∗∗ (0.008) −0.033∗∗ (0.014) −0.025∗∗∗ (0.004) −0.030∗∗∗ (0.008) −0.010∗ (0.005) −0.01 (0.008) −0.021∗ (0.012) −0.006∗ (0.004) −0.014∗ (0.008) −0.023∗∗∗ (0.006) −0.061∗∗∗ (0.010) −0.081∗∗∗ (0.015) −0.041∗∗∗ (0.004) −0.035∗∗∗ (0.010) N 861,698 300,156 861,698 861,698 861,698 300,156 861,698 861,698 861,698 Notes: Models include all variables in Table 3. Robust standard errors in parentheses. ∗signiﬁcant at 10%; ∗∗signiﬁcant at 5%; ∗∗∗signiﬁcant at 1%. Results for TFA in the third year should be interpreted with caution due to small sample sizes. H O W C H A N G E S I N E N T R Y R E Q U I R E M E N T S A L T E R T H E T E A C H E R W O R K F O R C E l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff two-stage model with three or fewer years of experience and decreases to .014 in the model with student ﬁxed effects. None of these estimates differ statistically from the estimates in table 3a. As in the base model, the estimated effects of temporary-license teachers are very similar to those of Teaching Fellows, and the effects of TFA members are very similar to those of college-recommended teachers. The speciﬁcation tests for ELA also give estimates that are similar to base model estimates. The Teaching Fellows and TFA members both appear to perform worse than college-recommended teachers. Temporary-license teach- ers fall between college-recommended and alternative route teachers. The difference between temporary and college-recommended and the difference between temporary and Teaching Fellows teachers are both statistically signif- icant ( p = .001 and p = .015, respectively). The results in tables 5a and 5b also allow us to assess the importance of controlling for aggregate student characteristics and including school ﬁxed effects. Excluding either school ﬁxed effects or student aggregates, but not both, gives results similar to the full model. However, the estimates are quite different when both aggregate student characteristics and school ﬁxed effects are omitted, even after controlling for detailed student-level characteristics. In particular, both Teaching Fellows and TFA members appear to do much worse when these are omitted. This result is not surprising. These teachers often teach in the most difﬁcult schools. Students in these schools are likely to perform more poorly than other students, even accounting for their prior scores and other individual characteristics. Without these controls, the esti- mates would attribute to the skills of these teachers’ school characteristics that differentially disadvantage students in the places where Fellows and TFA members teach. In addition to the base model and speciﬁcation checks presented in tables 5a and 5b, we estimated several other models. First, the prior models have as- sumed that pathways do not change in their characteristics over time, but this is not the case. Both Teaching Fellows and Teach for America, for example, continue to change both their selection criteria and their preservice training. Is there evidence that these programmatic changes affect student performance? To examine this, we estimated models by cohort. Unfortunately, the smaller sample sizes result in large standard errors when the sample is divided in this way. The strongest trend is an improvement in the math performance of Fel- lows over time. Second, teachers and students are mobile among classrooms, even within a school year. While we cannot fully account for this mobility in our estimates, we can estimate models that address this issue. When we limit our sample to teachers who taught in the same school from October through the date of the student achievement exam, we lose some observations, but the l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . . f f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 201 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE results are unaffected. Unfortunately, we do not have equivalent information for students. We can eliminate students who were not in the same school in the previous year. This removes all students who switched schools between the two academic years, including those who transitioned to middle school. The sample size for this analysis drops substantially, and we cannot identify the effects of Teaching Fellows or TFA in the middle school; however, the results for fourth and ﬁfth graders are very consistent with previous estimates. Third, some teachers may be more effective than others teaching students with certain characteristics. We explored this hypothesis for teacher pathways by estimating models that include interactions between pathway and class- room characteristics. However, except for evidence that college-recommended teachers do not do as well with classes that have students with a large range of prior test scores, there are few interactions worth noting.13 We ﬁnd that the effects of pathways differ by grade and thus estimate models that group students into two groups, a fourth- and ﬁfth-grade group and one composed of sixth, seventh, and eighth graders. The models are speciﬁed the same as the Pathway by Experience models in tables 3a and 5b. Table 6 summarizes the Pathway by Experience estimates. For elementary math, Teaching Fellows start out performing worse, but gain more over the ﬁrst year than do other teachers. By their second year, their students are doing as well as students of college-recommended teachers. TFA members also appear to make substantial gains during their ﬁrst year, but this coefﬁcient is never statistically different from that of a college-recommended teacher. For elementary ELA results, Teaching Fellows and TFA members initially have more trouble improving their students’ reading achievement than do temporary-license teachers or college-recommended teachers, and, unlike the situation for math, these teachers do not make differentially large gains in student achievement by their second year. By their third year, differences are not statistically signiﬁcant. The middle school estimates, especially for math, are quite different. While temporary-license teachers continue to see smaller gains than college- recommended teachers, Teaching Fellows do not perform signiﬁcantly worse than college-recommended teachers, on average, and TFA members initially perform better. In addition, Teaching Fellows make signiﬁcantly greater im- provements between their second and third year of teaching than do other teachers and appear to outperform both college-recommended and temporary- license teachers. Of the speciﬁcation checks, the one model with school- by-year-by-grade ﬁxed effects ﬁnds somewhat different results, with Fellows 13. All of these estimates are available from the authors on request. 202 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . f . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff Table 6. Effects of Teacher’s Pathway and Years of Experience on Student Achievement by Grade Pathway 1 Year 2 Years 3 Years Math grades 4, 5 College recommended 0 Individual evaluation NYC Teaching Fellows TFA Temporary license Other −0.009 −0.040∗∗∗ −0.034 −0.021 −0.049∗∗ 0.048 0.045 0.055 0.042 0.034 0.066 ELA grades 4, 5 College recommended 0 Individual evaluation 0.005 NYC Teaching Fellows −0.035∗∗∗ 0.035 0.031 0.006∗ 0.08 0.061 0.053 0.106 0.052∗∗ 0.059 0.054 0.074 0.039 TFA Temporary license Other −0.055∗ −0.015 −0.007 −0.015∗ −0.025 0.041 0.014 Math grades 6, 7, 8 0.047 0.008∗ 0.035 0.028 0.091∗ 0.113 0.04 0.056 0.025 0.017 College recommended 0 Individual evaluation NYC Teaching Fellows TFA Temporary license Other −0.02 −0.012 0.046∗∗ −0.018 −0.018 0.049 −0.017∗∗ 0.03 0.057 0.016∗∗ 0.021 College recommended 0 Individual evaluation NYC Teaching Fellows TFA Temporary license Other 0.013 −0.025∗ −0.030∗ −0.005 −0.021 ELA grades 6, 7, 8 0.032 0.032 −0.016∗∗ 0.114∗∗ 0.033 0.004 0.017 0.051 0.017 0.02 Notes: Models include all variables in table 3 and include the sam- ple identiﬁed in the table. All signiﬁcance tests are relative to college- recommended teachers with the same years of experience. ∗signiﬁcant at 10%; ∗∗signiﬁcant at 5%; ∗∗∗signiﬁcant at 1%. Results for TFA in the third year should be interpreted with caution due to small sample sizes. having a negative and signiﬁcant effect and TFA members having no average effect relative to college-recommended teachers. The results for sixth- through eighth-grade reading show negative point estimates for both Fellows and TFA teachers, but only the results for Fellows are statistically different from zero. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . . f f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 203 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE However, Fellows show a differential gain to the third year of experience and at that point outperform both college-recommended and temporary-license teachers. Table 6 allows an easy comparison of Teaching Fellows with college- recommended teachers; however, it is also interesting to see how Fellows perform relative to all other teachers in New York City. Because of the lower performance of teachers with temporary licenses or individual evaluation, the comparison of Fellows with all other teachers provides a more positive view of the gains of Fellows’ students. On average, there is no signiﬁcant difference between the gains of Fellows’ students and all others in grades 6–8, though elementary students do perform somewhat worse with ﬁrst-year Fellows than with all other ﬁrst-year teachers. However, in math second-year Fellows show a disproportionate gain and perform at least as well as all other second-year teachers. There is no evidence of such a gain for second- or third-year Fellows in ELA. How Important are the Pathway Effects? There are several ways of gaining perspective on the magnitude of the esti- mated effects of pathways. The ﬁrst is statistical signiﬁcance: do the estimates of alternative pathways and pathway interactions with experience differ from each other enough that we can be conﬁdent that the differences represent more than random ﬂuctuations? The estimates presented in tables 3 and 5 identify the pathway coefﬁcients that are statistically different from a ﬁrst-year college-recommended teacher; table 6 indicates the statistically different ef- fects when comparing a college-recommended teacher to teachers from other pathways with similar experience. A second means of gauging importance is to assess whether the esti- mated effects are big from a policy perspective. One way of assessing this is to examine the pathway effects relative to other effects. For example, a ﬁrst-year Teaching Fellow is estimated to show gains that are .02 lower in math than a ﬁrst-year college-recommended teacher (table 4). This effect is about 40 percent of the average difference in the improvement of students taught by second-year teachers relative to ﬁrst-year teachers (.02 compared to .05), which several other researchers have identiﬁed as important. Simi- larly, a Fellow on average has students who gain .03 points less in ELA than students of college-recommended teachers; this is approximately equivalent to 43 percent of the difference in gains between a student who is eligible for free lunch and one who is not. We estimate that a third-year Fellow teaching seventh-grade math improves student performance by 5.6 percent of a standard deviation more than a third-year college-recommended teacher, which is a bit closer but still less than the effect of free lunch 204 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff eligibility. Thus, relative to other estimated effects, pathway differences are not large. Alternatively, we could examine the relative increase in scale scores implied by the estimated pathway effects. For example, the average scale score of students on the 2002 seventh-grade math exam is about 670, with a standard deviation of 48. Our model estimates that the difference between a third-year Teaching Fellow and a third-year college-recommended teacher in seventh- grade math is .056 (.091 − .035). This translates into 2.7 scale points. At the median student score (671), an increase of 2.7 scale points moves a student in the middle of the test score distribution ahead of about 2.5 percent of the other students. Most would not judge this to be a large effect, but when this difference is accumulated over a few teachers over time, it could have a moderate impact. Teacher Improvement with Experience The estimated coefﬁcient of the interaction of pathway and experience com- bines several mechanisms. It includes the gains that individual teachers make over time, but it also includes differential attrition. If the best (worst) teach- ers leave during the ﬁrst year, the second-year estimates will look worse (better). To see whether selection or learning is driving these results, we estimate models that include ﬁxed effects for teachers. With teacher ﬁxed effects, we cannot identify the effects of pathways because an individual’s pathway does not change over time. However, we can include pathway by experience interactions. We ﬁnd no statistically signiﬁcant difference in the returns to experience across pathways. The few differences in path- way returns to experience shown in table 6 are likely due to differences in attrition. In addition to estimating the previous models, we examine whether the effectiveness of pathways differs by the attributes of students and whether successive cohorts of teachers yield different gains in student achievement. We hypothesized that different pathways may be relatively more effective with students with differing characteristics, such as prior academic achieve- ment or race. We estimate models that include interactions between pathway and classroom characteristics and found little evidence of this by pathway. Likewise, we hypothesized that as the attributes of pathways change over time, they may have differing effects on student achievement. Both Teach- ing Fellows and TFA, for example, continue to change both their selection criteria and their preservice training. Is there evidence that these program- matic changes affect student performance? Unfortunately, the sample sizes get small, and the standard errors large, when the sample is divided in this way. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 205 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE The strongest trend is an improvement in the math performance of Fellows over time.14 Differential Turnover and the Overall Effect of Pathways The analyses do not present a simple story, but a few important relationships emerge. First, it is clear from the numbers in table 1 that the composition and overall supply of teachers changes as a result of changes to certiﬁcation requirements; in opening up new certiﬁed pathways having lower preser- vice requirements and subsidized education, uncertiﬁed teachers were largely eliminated. Many teachers entering through these new pathways have strong academic training and perform well on tests of general knowledge. How- ever, there is some evidence that teachers that enter New York City elemen- tary schools through these pathways do not teach mathematics or ELA as well as teachers from the college-recommended path during their ﬁrst year. Their mathematics teaching does improve differentially between their ﬁrst and second year, and they are approximately equivalent to college-recommended teachers during their second year. Teachers who enter New York City middle schools through these pathways do better than those who teach in elemen- tary schools. Teach for America members do a better job of teaching math- ematics to middle school students than either temporary-license or college- recommended teachers during their ﬁrst year, while Fellows perform at least as well as these teachers. By their third year of experience, Fellows appear to outperform college-recommended and temporary-license teachers. The re- sults for ELA in middle school are not as strong. The students of both Fellows and TFA teachers in their ﬁrst year have somewhat lower achievement gains than college-recommended or temporary-license teachers. Teaching Fellows show differentially strong improvement in teaching between their second and third years. Fellows with three years of experience perform about as well as college-recommended teachers with three years of experience and exceed the performance of temporary-license teachers. These estimates tell us something about the relative effectiveness of the pathways for ﬁrst-year, second-year, and third-year teachers, but they do not incorporate the differential retention of these teachers and thus do not reﬂect the average quality of teachers that we might expect to get if we ﬁlled teach- ing positions with teachers from certain pathways relative to other pathways. How would the academic gains of students differ if school ofﬁcials systematically ﬁlled job openings by hiring teachers entering through one pathway versus another? The answer in part depends upon the relatively effectiveness of the ﬁrst-year, 14. Estimates for these models can be found in “How changes in entry requirements alter the teacher workforce and affect student achievement” at www.teacherpolicyresearch.org. 206 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d f . . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff Table 7. Cumulative Teacher Attrition Rates by Pathway for Elementary, Middle, and Junior High School Teachers in New York City, 2000–2004 Experience College recommended Individual evaluation Teaching Fellows Teach for America Temporary license Other 1 2 3 4 1 2 3 4 0.115 0.212 0.290 0.368 0.143 0.261 0.354 0.446 0.139 0.256 0.322 0.391 Actual 0.105 0.278 0.434 0.544 0.107 0.477 0.727 0.850 0.184 0.300 0.413 0.501 Adjusting for differences between school environments 0.147 0.276 0.355 0.438 0.096 0.265 0.433 0.547 0.088 0.426 0.667 0.813 0.167 0.327 0.435 0.519 0.264 0.402 0.500 0.573 0.252 0.396 0.510 0.592 second-year, and third-year teachers differentiated by pathway discussed above. However, it is also necessary to account for differences in retention rates across pathways. This follows from the meaningful gains in teacher value-added mea- sures associated with increased experience over the ﬁrst few years of teaching. If one pathway consistently has higher turnover, even if its teachers do well relative to those in other pathways with the same experience, the pathway may not, on average, be providing the most effective teachers. Descriptive statistics characterizing the attrition rates for teachers by path- way in elementary, middle, and junior high schools are shown in table 7. The statistics reﬂect the experiences of the cohorts of teachers who began teaching in New York City from 1999–2000 to 2003–4. The table shows that 11.5 per- cent of college-recommended teachers leave teaching after the ﬁrst year and that after four years 37 percent have left. The turnover rates are higher for the temporary-license group and the other groups in the ﬁrst year and are substantially higher for TFA members in the following years.15 Interpreting the meaningful differences in attrition rates shown is compli- cated by several factors. With the more recent cohorts of entering teachers not having been observed for the full ﬁve years, the attrition rates for greater levels of experience only reﬂect the pattern for the earlier entering cohorts. This is problematic in that a closer look at the underlying data reveals that attrition rates generally are higher in some years than others. A related issue is that teachers entering through alternative pathways disproportionately entered in recent years. As a result, experience effects could mistakenly be attributed 15. Note that teachers who transfer to positions in public charters and nonteaching positions are counted as leaving. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . f . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 207 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE to experience, pathways, or a combination of the two. Interpretation is also complicated by the fact that the attrition rates of teachers from various path- ways may in part reﬂect differences in the kinds of schools they are assigned. To account for these and other issues, we estimate a conditional logit model predicting whether or not a teacher leaves teaching in New York City the fol- lowing year accounting for pathway, experience dummies, the interaction of experience and pathway, year ﬁxed effects, and school ﬁxed effects. Thus, these estimates are based upon within-school differences in teacher attrition rates and hold constant any effects that vary between schools. The results indicate that the differences in attrition rates across pathways are statistically signiﬁcant and meaningful in magnitude. For example, 9.6 percent of Teaching Fellows are predicted to leave teaching after the ﬁrst year, substantially less than the attrition of college-recommended teachers. The difference between the alternative routes and college-recommended teachers is signiﬁcant but not as great as prior to controlling for school ﬁxed effects. Over time the relative attrition of Teaching Fellows grows so that it is roughly comparable to college-recommended teachers after two years and exceeds them after three and four years. Those entering through TFA also are relatively more likely to return in the second year. The retention of TFA teachers beyond the second year falls off dramatically, relative both to the traditional routes and to the Fellows pathway. After four years, fewer than 20 percent of TFA teachers are predicted to remain teaching in New York City public schools. How does the average-value-added measure of teachers vary across path- ways once differences in teacher retention rates are taken into account? We address this question using the following simulation. Suppose that school ofﬁ- cials hired an arbitrary number of new teachers (e.g., one thousand) from each of the pathways. For subsequent years, the teachers hired from each pathway are allowed to age through the experience distribution, applying the pathway- dependent retention rates implied by the conditional logit model. Teachers who quit are replaced by new teachers from the same pathway. These new hires in turn age through the system. In this way, it is possible to simulate how the experience distribution of teachers from each pathway would evolve over time and differ across pathways and, in turn, simulate how such dif- ferences would affect the average-value-added measures of the teachers from each pathway.16 Simulation results are shown in table 8 for the value-added ELA and Math models estimated separately for grades 4 and 5 and 6–8. For comparison 16. Given the small number of TFA members and Fellows with greater than three years of experience, we assume that after three years of experience, student achievement gains remain equal and constant across pathways and the retention differences remain constant and reduce at the overall quit rate for years four and higher. 208 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 Table 8. Predicted Average Value Added by Pathway, Grade Groupings, and Years ELA 4&5: AVERAGE VA BY PATHWAY AND EXPERIENCE SINCE IMPLEMENTATION MATH 4&5: AVERAGE VA BY PATHWAY AND EXPERIENCE SINCE IMPLEMENTATION Simulation year CR IE TF TFA TL Other Simulation year 1 2 3 4 5 0.000 0.005 −0.035 −0.055 −0.015 −0.007 0.030 0.027 0.002 −0.019 0.032 0.009 0.044 0.058 0.023 −0.035 0.035 0.006 0.054 0.069 0.027 −0.026 0.044 0.013 0.059 0.076 0.029 −0.027 0.048 0.017 1 2 3 4 5 CR IE TF TFA TL Other 0.000 −0.009 −0.040 −0.034 −0.021 −0.049 0.041 0.037 0.046 0.035 0.025 0.037 0.065 0.048 0.037 0.052 0.036 0.038 0.071 0.056 0.039 0.046 0.042 0.042 0.073 0.058 0.041 0.048 0.043 0.044 VA BY PATHWAY AND EXPERIENCE VA BY PATHWAY AND EXPERIENCE Experience CR IE TF TFA TL Other CR IE TF TFA TL Other 1st year 0 0.005 −0.035 −0.055 −0.015 −0.007 1st year 0.000 −0.009 −0.040 −0.034 −0.021 −0.049 2nd year 0.035 0.031 0.006 −0.015 0.041 0.014 2nd year 0.048 0.045 0.055 0.042 0.034 0.066 3rd year 4th year 5th Year 0.054 0.074 0.039 −0.025 0.047 0.008 3rd year 0.080 0.061 0.053 0.106 0.052 0.059 0.068 0.088 0.053 −0.011 0.061 0.022 4th year 0.088 0.069 0.061 0.114 0.060 0.067 0.077 0.097 0.062 −0.002 0.07 0.031 5th Year 0.091 0.072 0.064 0.117 0.063 0.070 D . B o y d , P. G r o s s m a n , H . L a n k f o r d , S . L o e b , a n d J . W y c k o f f 2 0 9 l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . f . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 2 1 0 I E D U C A T O N F I N A N C E A N D P O L I C Y Table 8. Continued. ELA 6-8: AVERAGE VA BY PATHWAY AND EXPERIENCE SINCE IMPLEMENTATION MATH 6-8: AVERAGE VA BY PATHWAY AND EXPERIENCE SINCE IMPLEMENTATION CR 0 IE TF TFA TL Other CR IE TF TFA TL Other 1 2 3 4 5 0.013 −0.025 −0.03 −0.005 −0.021 0.027 0.029 −0.017 0.027 0.002 0.007 0.022 0.018 0.078 0.021 0.011 0.011 0.017 0.013 0.061 0.018 0.007 0.007 0.025 0.020 0.065 0.022 0.013 0.012 1 2 3 4 5 0.000 −0.020 −0.012 0.046 −0.018 −0.018 0.042 −0.017 0.026 0.056 0.010 0.011 0.032 0.015 0.067 0.085 0.026 0.034 0.044 0.029 0.071 0.081 0.037 0.045 0.047 0.033 0.071 0.082 0.040 0.048 VA BY PATHWAY AND EXPERIENCE VA BY PATHWAY AND EXPERIENCE CR IE TF TFA TL Other CR IE TF TFA TL Other 1st year 0.000 0.013 −0.025 −0.030 −0.005 −0.021 1st year 0.000 −0.020 −0.012 0.046 −0.018 −0.018 2nd year 0.032 0.032 −0.016 0.033 0.004 0.017 2nd year 0.049 −0.017 0.030 0.057 0.016 0.021 3rd year 0.025 0.017 0.114 0.051 0.017 0.020 3rd year 0.035 0.028 0.091 0.113 0.040 0.056 4th year 0.017 0.009 0.106 0.043 0.009 0.012 4th year 0.054 0.047 0.110 0.132 0.059 0.075 5th Year 0.030 0.022 0.119 0.056 0.022 0.025 5th Year 0.059 0.052 0.115 0.137 0.064 0.080 H O W C H A N G E S I N E N T R Y R E Q U I R E M E N T S A L T E R T H E T E A C H E R W O R K F O R C E l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . f . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff purposes, the value-added estimates by pathway and experience level are also shown below each of the simulations by grade level. As an example, consider teachers entering via TFA and teaching fourth- or ﬁfth-grade math. Their average value added in the ﬁrst year of the simulation (relative to ﬁrst-year college-recommended teachers) is estimated to be −0.034, merely reﬂecting the estimated value added for a ﬁrst-year TFA teacher and the fact that all the teachers are in their ﬁrst year. In the second year of the simulation, 91.2 percent of the teachers would be in their second year, with the remainder newly hired, implying an average value added of 0.035, a ﬁgure lower than the 0.042 for a second-year TFA math teacher in fourth or ﬁfth grade but higher than the −0.034 estimated for such a teacher in the ﬁrst year. Rolling the simulation forward, it is possible to see how the average-value-added measure for a particular pathway would change over time as well as how the average- value-added estimates would differ across pathways after any given number of years. In general, pathways with high teacher attrition will have their overall student achievement gains reduced as inexperienced teachers with lower stu- dent achievement gains are substituted for teachers who would have produced stronger gains in student achievement. Again, the average-value-added results do not tell a simple story. It is still the case that the Teaching Fellows and TFA perform well relative to other pathways in middle school math. In elementary math, after a somewhat weaker ﬁrst year, they do approximately as well as teachers from all other routes except for college-recommended. For middle school ELA, the Fellows start out somewhat worse but outperform other routes in later years. TFA teachers also start somewhat worse but roughly catch up in later years. In fourth- and ﬁfth-grade ELA, the Fellows perform somewhat worse than other routes, with the exception of TFA teachers, who perform worse than Fellows. 6. CONCLUSIONS We ﬁnd that there are relatively small differences in student achievement improvement attributable to preparation pathways, and these effects typically exist only when comparing ﬁrst-year teachers. What might be the story behind these results? One possibility is that the differential effectiveness of teachers from different pathways comes from differences in experiences working with students prior to beginning as the teacher of record. College-recommended teachers can spend hundreds of hours in the classroom before taking control of their own class, while alternative-route teachers may spend comparatively little time in classrooms due to the short duration of their preservice prepa- ration. A second possible cause for the differences in effectiveness may be differences in coursework prior to entry. This coursework may cover topics l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . . . . f f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 211 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE from child development and classroom management to how to teach read- ing and mathematics. In general, Teaching Fellows and TFA teachers have stronger math backgrounds than teachers from other pathways. Some would suggest that teaching middle school math draws on this stronger background more directly, yielding these teachers a relative advantage in these subjects. In addition, Teaching Fellows and Teach for America members, as well as other Transitional B teachers, enroll in coursework during their ﬁrst two years of teaching. Our teacher ﬁxed effects estimates indicate that the differential ef- fects of experience across pathways is probably due to the differential attrition of weaker teachers across pathways, not to improvements in performance of individual teachers. The administrative data employed here is not well suited to understanding the variety of selection and preparation differences that exist across and within paths. We are in the midst of what amounts to a national experiment in how best to attract, prepare, and retain teachers, particularly for high-poverty urban schools. New York City offers a valuable setting in which to study the ef- fects of different pathways into teaching, as a number of pathways coexist in the same labor market. In many respects, alternative routes are designed as a recruitment strategy. By targeting elite colleges and universities or re- cruiting career-changers, both TFA and Teaching Fellows succeed in attract- ing, and selecting, teachers with stronger academic backgrounds, at least as measured by selectivity of undergraduate institution and test scores. In large measure alternative-route teachers are replacing teachers who previously had been uncertiﬁed and thus had not necessarily had any prior or concur- rent preparation for teaching. Alternate-route teachers are typically placed in schools with more students of poverty, schools that generally have the hardest time attracting and retaining teachers. Both the Fellows and TFA programs are designed speciﬁcally to place teachers in underserved or difﬁcult-to-staff schools. Our analysis of alternative-route teachers suggests that in some instances Fellows and TFA members provide higher student achievement gains than the temporary-license teachers they replace. For example, Fellows in their third year of teaching in middle schools outperform temporary-license teach- ers in both math and ELA. More typically, alternative-route teachers are no better or worse than the temporary-license teachers they replace. When com- pared to college-recommended teachers, alternative-route teachers often pro- vide smaller gains in student achievement, at least initially, and for ELA it takes longer to catch up. As noted above, many of these differences are not large in magnitude, typically about 2 to 5 percent of a standard deviation, and the variation in effectiveness within pathways is far greater than the average differences between pathways. 212 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . f . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff In the end, this is not likely to be a story about which pathway is best, nor should it be. For example, without alternative routes to teaching it is unlikely New York City could meet New York State and federal requirements to have a qualiﬁed teacher in every classroom. Based on this analysis we believe that different pathways bring different strengths to teaching. These differences in pathways provide a means to discuss potential improvements in how all pathways prepare teachers. They also suggest that the assignment of teachers to grades and subjects may be inﬂuenced by pathway. The current analysis does not address a number of important policy ques- tions. For example, what attributes of preparation are most important in in- creasing student achievement. One can imagine that the gross aggregations by pathways mask the potential strong effects of speciﬁc preparation attributes. We are in the process of collecting additional data toward this goal, as well as hoping to shed light on which characteristics of pathways and preservice edu- cation (not simply which pathways) affect teachers’ ability to enhance student learning. As part of this data collection effort, we surveyed program participants as they completed their teacher education programs in the spring of 2004 at seventeen institutions that provide many of the new college-recommended teachers to New York City. In addition, we surveyed all Teaching Fellows and Teach for America members who in 2004 were completing their summer training to prepare them to take jobs in New York City schools. On this sur- vey we asked questions about the participants’ family background, education, selection of pathway into teaching, and many facets of their preparation. In addition, we have completed a survey of all six thousand ﬁrst-year teachers in New York City public schools inquiring about their background, prepa- ration, and induction and the environment in which they work. Finally, we are just completing a detailed analysis of the preparation programs at sev- enteen schools of education and alternative route programs to understand the structure, requirements, and content of preparation that teachers receive. This information will be used to identify the attributes of teacher preparation that are most inﬂuential in the achievement of students and the retention of teachers. Another interesting question is how the size and composition of the college- recommended pool of teachers would change if these teachers received the ﬁnancial incentive of reduced tuition toward a master’s degree, as is cur- rently offered by the alternative-route programs. Would they be able to at- tract a stronger pool of applicants, and would that result in higher student achievement gains for that path? This is more difﬁcult to assess. Given the demand for new teachers, debates over entrance requirements, recruitment strategies, and teacher preparation are likely to continue for some time. How- ever, the quality of such debates should improve as data from large-scale l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . f . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 213 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE investigations such as this one begin to inform policy decisions about entry into teaching. We are grateful for data provided by the New York City Department of Education and the New York State Education Department. We also appreciate helpful comments regarding data from Vicki Bernstein, Tony Chu, Ron Danforth, Eddie Mauro, Lori Mei, Joseph Porter, Abigail Smith, and Harvey Weinstein. We also appreciate help- ful comments on an earlier draft from Elizabeth Arons, Carla Asher, Vicki Bernstein, Marilyn Cochrane Smith, Johanna Duncan Poitier, Ronald Ehrenberg, Robert Flo- den, Joseph Frey, Nicholas Michelli, Ruth Pagery, Ana Maria Villegas, and Nancy Willy-Schiff. In addition, we appreciate comments from David Figlio and two anony- mous reviewers. This research is being supported by funding from the City University of New York (through a grant from the Corporation for National and Community Ser- vice under an AmeriCorps grant), the Carnegie Corporation of New York, the National Science Foundation (REC-0337061), the New York State Department of Education, and the Spencer Foundation. The views expressed in this article are solely those of the authors and do not necessarily reﬂect those of any of the funders. Any errors are the responsibility of the authors. REFERENCES Ball, D. L., and D. Cohen. 1999. Developing practice, developing practitioners. In Teaching as the learning profession: Handbook of policy and practice, edited by L. Darling- Hammond and G. Sykes, pp. 3–32. San Francisco: Jossey Bass. Ballou, Dale. 2002. Sizing up test scores. Education Next 2 (Summer): 10–15. Boyd, D., H. Lankford, S. Loeb, and J. Wyckoff. 2005. The draw of home, how teachers’ preferences for proximity disadvantage urban schools. Journal of Policy Analysis and Management 24 (1): 113–34. Brown, C. A., and H. Borko. 1992. Becoming a mathematics teacher. In Handbook of research on mathematics teaching and learning, edited by D. A. Grouws, pp. 209–39. New York: Macmillan. Carnegie Forum on Education and the Economy. 1986. A nation prepared: Teachers for the 21st century. New York: Carnegie Forum on Education and the Economy. Cochran-Smith, M., and K. Zeichner, eds. 2005. Review of research in teacher education. Washington, DC: American Educational Research Association. Darling-Hammond, L., D. J. Holtzman, S. J. Gatlin, and J. V. Heilig. 2005. Does teacher preparation matter? Evidence about teacher certiﬁcation, Teach for America, and teacher effectiveness. Working paper, Stanford University. Darling-Hammond, L., ed. 2000. Studies of excellence in teacher education. Washington, DC: American Association of Colleges of Teacher Education. Darling-Hammond, L., and M. McLaughlin. 1999. Investing in teaching as a learning profession. In Teaching as the learning profession: Handbook of policy and practice, edited by L. Darling-Hammond and G. Sykes, pp. 263–91. San Francisco: Jossey Bass. 214 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . f . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 D. Boyd, P. Grossman, H. Lankford, S. Loeb, and J. Wyckoff Decker, Paul T., Daniel P. Mayer, and Steven Glazerman. 2004. The effects of Teach for America on students: Findings from a national evaluation. Report for Mathematica Policy Research. Ehrenberg, R. G., and D. J. Brewer. 1995. Did teachers’ verbal ability and race matter in the 1960s: Coleman revisited. Economics of Education Review 14 (1): 1–23. Feiman-Nemser, S. 1990. Teacher preparation: Structural and conceptual alternatives. In Handbook of research on teacher education, edited by W. R. Houston, pp. 195–211. New York: Macmillan. Feiman-Nemser, S. 1983. Learning to teach. In Handbook of teaching and policy, edited by L. Shulman and G. Sykes, pp. 150–70. New York: Longman. Finn, Chester E., Jr., and Kathleen Madigan. 2001. Removing the barriers for teacher candidates. Educational Leadership 58 (8): 29–31. Garet, M. S., A. C. Porter, L. Desimone, B. F. Birman, and K. S. Yoon. 2001. What makes professional development effective? Results from a national sample of teachers. American Educational Research Journal 38 (4): 915–45. Goodlad, J. I. 1990. Teachers for our nation’s schools. San Francisco: Jossey Bass. Holmes Group. 1986. Tomorrow’s teachers. East Lansing, MI: The Holmes Group. Kane, Thomas J., and Douglas O. Staiger. 2001. Improving school accountability measures. NBER Working Paper No. 8156. Lankford, H., S. Loeb, and J. Wyckoff. 2002. Teacher sorting and the plight of urban schools: A descriptive analysis. Educational Evaluation and Policy Analysis 24 (1): 38–62. Loucks-Horsley, S., and C. Matsumoto. 1999. Research on professional development for teachers of mathematics and science: The state of the scene. School Science and Mathematics 99 (5): 258–71. McKibbin, M. D. 1998. Voices and views: Perspectives on California’s teaching and internship programs. Sacramento, CA: Commission on Teacher Credentialing. Monk, D. H. 1994. Subject area preparation of secondary mathematics and science teachers and student achievement. American Educational Research Journal 28:237– 74. National Commission on Teaching and America’s Future. 1996. What matters most: Teaching for America’s future. Woodbridge, VA: National Commission on Teaching and America’s Future. National Research Council. 2001. Educating teachers of science, math, and technology: New practices for the new millennium. Washington, DC: National Academy Press. New York City Teaching Fellows. 2005. NYCTF. Accessed 21 November 2005. Available http://www.nyctf.org. Raymond, Margaret, Stephen H. Fletcher, and Javier Luque. 2001. Teach for America: An evaluation of teacher differences and student outcomes in Houston, Texas. CREDO Report, The Hoover Institute, Stanford University. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . . f . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 215 HOW CHANGES IN ENTRY REQUIREMENTS ALTER THE TEACHER WORKFORCE Rivkin, S. G., E. A. Hanushek, and J. F. Kain. 2005. Teachers, schools, and academic achievement. Econometrica 73 (2): 417–58. Rockoff, Jonah. 2004. The impact of individual teachers on student achievement: Evidence from panel data. American Economic Review 94 (2): 247–52. Rogosa, David. 2002. Irrelevance of reliability coefﬁcients to accountability systems: Statisti- cal disconnect in Kane-Staiger “Volatility in School Test Scores.” Working Paper, Stanford University. Sanders, William. 2003. Correspondence: Margin of error. Education Next 3 (Winter): 4. Sanders, W. L., and S. P. Horn. 1994. The Tennessee Value-Added Assessment System (TVAAS): Mixed-model methodology in educational assessment. Journal of Personnel Evaluation in Education 8:299–311. Sanders, W. L., and J. C. Rivers. 1996. Research project report: Cumulative and residual effects of teachers on future student academic achievement. University of Tennessee, Value-Added Research and Assessment Center. Shen, Jianping. 1999. Alternative certiﬁcation: Math and science teachers. Educational Horizons 78 (1): 44–48. U.S. Department of Education. 2002. Meeting the highly qualiﬁed teachers challenge. U.S. Department of Education, Ofﬁce of Postsecondary Education, Ofﬁce of Policy Planning and Innovation. Wilson, S., R. Floden, and J. Ferrini-Mundy. 2002. Teacher preparation research: Current knowledge, gaps, and recommendations. University of Washington, Working Paper. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 1 2 1 7 6 1 6 8 8 9 6 9 e d p 2 0 0 6 1 2 1 7 6 p d . f . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 216 EDUCATION FINANCE AND POLICY
下载pdf